The Attention Economy

The Milken Institute Review • Third Quarter 2015 • volume 17, number 3
In this issue
claudia goldin
On closing the gender pay gap.
The key is flexible work schedules.
larry fisher
On hydrogen-powered cars.
Miracles sometimes happen.
frank rose
On the Internet battle for eyeballs.
Time is the scarce resource.
pallavi aiyar and chin hwee tan
On comparing the Indian and Indonesian economies.
More alike than you think.
thomas healey and catherine reilly
A Journal Of Economic Policy
On the coming global pension disaster.
Pay now or pay later.
tomas philipson
On financial engineering for pharmaceutical R&D.
Managing risk with derivatives.
marsha vande berg
allen sanderson and john siegfried
On the NCAA monopoly.
Pay players what they’re worth.
bill frey
On neighborhood desegregation.
The bonus from diversity.
A dove on inflation.
On China’s high-wire act.
Reform or growth – must they choose?
Se
th
co n d
s Co u
t
n
e h
igh
t le io n
t
a
f
b
r
t
fo
r y o e q u e n c y atte n
u r c y be r-
the attention economy
by f ran k rose
Nearly a half-century ago, in a lecture
sponsored by Johns Hopkins University
and the Brookings Institution, the economist and future Nobel laureate Herbert A.
Simon told a story about bunny rabbits. It
seems his neighbors had purchased a pair
of bunnies for their daughter as an Easter
present, and since the rabbits were of different genders the neighbors soon found
themselves living in, as Simon put it, “a
rabbit-rich world.” This would have consequences not just for the neighbors but for
the local food supply. “A rabbit-rich world
is a lettuce-poor world,” Simon pointed out,
“and vice versa.”
42
The Milken Institute Review
Alas for leporidologists, Simon was not
giving a disquisition on rabbits. His actual
topic was information, which even then appeared to be exploding, and he went on to
make an observation that has been cited
many times since:
In an information-rich world, the wealth of
information means a dearth of something
else: a scarcity of whatever it is that information consumes. What information consumes
is rather obvious: it consumes the attention of
its recipients. Hence, a wealth of information
creates a poverty of attention and a need to
allocate that attention efficiently among the
overabundance of information sources that
might consume it.
Third Quarter 2015
43
tk kotryna zukauskaite
th e atte ntion economy
And so, at the dawn of the information age,
was born the notion that we live in an “attention economy” in which a glut of information
leaves us with a deficit of attention. It was a
radical idea, since for most of human history
it has been information that’s in short supply
and attention that’s abundant. But Simon was
prescient. In a few quick sentences, he predicted a reversal of the economic relationship
between media producers and media consumers. In the future, the value of information
(the stuff being produced) would trend toward zero, while the value of attention, which
is owned by consumers but can be leveraged
by companies that help them allocate it, would
only rise. Google, its founders as yet unborn,
would triumph; newspapers would collapse.
Of course, it wasn’t quite that simple. There
were other issues to be decided, chief among
them the question of how to measure this
newly identified resource called attention.
Simon thought the answer was fairly obvious:
attention should be measured by the amount
of time an average business executive, a person
he identified as having a bachelor’s degree and
an IQ of 120, spends focused on something.
Such “attention units” would capture the cost,
in addition to any monetary outlay, of receiving information.
But Simon was a good 30 years ahead of his
time. By the time the rest of the world caught
up with his ideas, other, cruder, metrics had
come into use – metrics that have so distorted
the economics of the Internet that we find ourselves awash in information that’s useless, even
predatory, while information that actually deserves our attention often goes begging. Fixing
F R AN K ROS E is a senior fellow at the Columbia
University School of the Arts and the author of The Art
of Immersion: How the Digital Generation Is Remaking
Hollywood, Madison Avenue and the Way We Tell Stories.
44
The Milken Institute Review
this won’t be easy, but it’s going to be critical
to a functioning media industry – as a growing number of people are starting to point out.
mau-mauing the flak catchers
Last January, Evan Williams became the latest
to speak up. Williams is a co-founder of Twitter, the microblogging service that went public in 2013, and more recently of Medium, an
online publication that doubles as a blogging
platform and has serious journalistic ambitions. In December, Facebook announced
that Instagram – the popular photo-sharing
app that it bought for $1 billion in 2012 – had
300 million “monthly active users” (MAUs, in
the lingo of the trade). This was remarkable
for a service that nine months earlier had
only 200 million.
But commentators on Wall Street and in
the media immediately ginned up a comparison with Twitter, whose user base, after an
initial growth spurt, stood a little below 300
million – much to the Street’s distress. From
CNBC to Mashable to Adweek, one outlet
after another proclaimed Instagram to be
“bigger than Twitter.” A Citibank analyst announced that Facebook’s not-yet-profitable
photo app might be worth $35 billion, far
more than Twitter’s market cap of $24 billion.
A columnist for CNNMoney opined that
Twitter should sell itself to the highest bidder.
Williams’s response, delivered to a Fortune
reporter: “I don’t give a s***.” And for good
reason. Aside from the fact that they’re interactive services that live on Internet-connected
electronic devices, Twitter and Instagram
have little in common. Twitter, as Williams
pointed out, is a “realtime information network” where big news breaks first and world
leaders and celebrities speak to global audiences. But while Twitter can be highly addictive to initiates, it has not made itself friendly
to newbies – a failing that contributed to in-
vestors’ feelings of relief when its CEO resigned in June. Instagram, on the other hand,
is a fun and extremely well-executed app that
encourages people to connect over photos.
Monthly active users can be a helpful yardstick for such online services, but as Slate’s
Will Oremus pointed out in one of the few informed appraisals of the Twitter/Instagram
contretemps, “they aren’t the only one. Others might include the amount of time users
traffic wins, regardless of how fleeting that
traffic might be.
We’ve been here before, of course. “Uniques”
are the online equivalent of ratings on television. The fixation on ratings fed the lowestcommon-denominator effect that held the industry in its grip from the late 1950s until just
a few years ago, when the rise of pay-TV channels and the growing sophistication of audiences pushed television into a new
55 percent of the people who visit a Web
page stay for less than 15 seconds. Yet the site
with the most traffic wins, regardless of how
fleeting that traffic might be.
spend on the network, the amount of content
they post, and the number of people who see
that content.” By those measures, Twitter far
outstrips its so-called rival: 500 million tweets
per day compared to 70 million photos posted
to Instagram; 500 million people per month
who visit the site but don’t log in (and therefore aren’t counted as “active users”); 185 billion impressions per quarter.
In Twitter’s case, a shortage of MAUs
turned off advertisers, weighed on the stock
price, and helped precipitate a change in leadership. But as misleading as the MAU metric
can be for social sites, the monthly tally of
“unique visitors” – typically used to gauge the
importance of media sites – is worse, for the
stock price and for ad rates alike. It assumes
that if someone lands on a Web page, that
person is going to stay there long enough to
read or watch whatever is onscreen. The reality is quite different. According to the Web
analytics firm Chartbeat, 55 percent of the
people who visit a Web page stay there for less
than 15 seconds. Yet the site with the most
“golden age” typified by serials like
Mad Men and Girls. But the situation is
even worse online because only a third of
Web advertising is bought on the same costper-thousand basis that prevails on TV. The
rest is “performance based,” meaning that advertisers pay a fraction of a penny each time
someone clicks on their ads. So it’s not enough
for advertisers to reach as many eyeballs as
possible; in order for those eyeballs to count,
they have to generate some sort of response.
That makes sense for Google, which generates the great bulk of its outsized profits ($14.4
billion on revenues last year of $66 billion)
by serving up precisely targeted ads next to
search results. But it has fed the perception in
the ad business that online display advertising
– banner ads that run across the top of a Web
page, for instance – is ineffective because it
doesn’t generate as many clicks as search advertising. So rather than being treated as a
brand-building medium like television, the
Internet has devolved into a direct-response
medium – a bargain-basement ad emporium
Third Quarter 2015
45
tk
46
The Milken Institute Review
that’s only as good as the next click. Which will
be a problem since the Internet is where television is headed.
It doesn’t help that the Internet is essentially unaffected by physical constraints. Television, especially analog television, has discernible limits: there are only so many viewing
hours in the day, only so many megahertz of
bandwidth that can be devoted to broadcasting channels simultaneously, only so many
minutes that can be devoted to ads without
audiences tuning out completely. (The current rule of thumb in the United States is 8
minutes out of every 30.) Not so online, where
the potential number of ad-carrying Web
pages is effectively infinite.
At the same time, online distribution is
largely controlled by users rather than publishers, either through search engine queries
or by spreading information virally on sites
like Facebook and Twitter. Marc Andresson, a
leading Silicon Valley venture capitalist, has
been talking about “a golden age of journalism” that could follow a transition to a newgeneration audience that's mobile and always
connected. But with users driving distribution
and with infinite inventory pushing ad rates
relentlessly downward, the focus on unique
visitors and click-through rates has inspired a
lot of Web publishers to try to game the system – to go big by providing junk content to
go with the junk ad medium that serves it up.
kotryna zukauskaite
what the algorithm said to the
content farm
The first target for abuse was search. Through
the miracle of search engine optimization,
Web publishers can design their sites to attract notice from the bots that crawl the Net
on behalf of Google and other search engines.
Figure out how to take maximum advantage
of this while delivering minimal value and
you get something like Demand Media, the
“content farm” that five years ago looked like
the future of journalism.
And a dismal future it would have been: the
idea was to pay writers and videographers a
pittance – $15 or so – to churn out near-useless
material on topics a computer algorithm said
people wanted to know about, then lard it up
with ads and rely on search engines to drive
traffic. Most visitors would go straight back to
Google, but who cared? A unique was a
unique, no matter how fleeting the visit, and
Demand Media was soon getting more than
100 million uniques a month, making it one of
the top 20 Web properties in the United States.
Demand Media’s business model was stunningly cynical, though the company did try to
dress it up with self-serving rhetoric about
“publishing what the world wants to know
and share.” More surprising was the number
of tech-savvy individuals who bought in.
“They really understand consumer behavior
on the Web and how to build businesses on it,”
Facebook COO Sheryl Sandberg said to
Bloomberg Businessweek. Wired magazine
concluded at the end of a lengthy profile that
“the Demand way may be inescapable.”
Investors certainly seemed to think so. In
January 2011, when Demand went public in
an IPO led by Goldman Sachs, the market
valued it at $1.9 billion on opening day. Never
mind that it had lost money in each of its four
years of existence; it was the first IPO to top
$1 billion since Google itself went public in
2004. The New York Times Company, meanwhile, was valued at $1.55 billion.
That was then. By November 2013, not
quite three years later, Demand Media’s traffic had fallen by half, its CEO had resigned,
and its stock price had dropped from a peak
of $52.50 shortly after the IPO to somewhere
around $5. What happened? Google.
Days before Demand’s initial public offering, Google’s engineer in charge of fighting
Third Quarter 2015
47
th e atte ntion economy
Web spam noted in a company blog post that
users wanted it to take action against content
farms and the like. His advisory was largely
ignored in the froth of the IPO. But a few
weeks later, Google made good on the threat
by introducing a significant change in the way
it tallied search results. Its new search algorithm, called Panda, specifically penalized
low-quality sites – those with thin content
and too many ads.
Demand Media’s traffic plummeted. By
April 2011, outside analytic services were reporting that visits to its sites were down as
much as 40 percent. The stock price started
falling accordingly. Copycat sites, of which
there was no shortage, suffered a similar fate.
Google “wouldn’t give us any relief,” one competitor told an interviewer from Harvard’s
Nieman Journalism Lab, “so I realized this
was not a sustainable business.”
With search out, would-be media innovators turned to social. You would think that
sharing through social media would be relatively immune to the gamesmanship that corrupted search, since for something to go viral
it presumably has to deliver on some level.
(There’s a reason that homemade cat videos
tend to be insanely popular.) But that turned
out not to be the case. No sooner did content
farms implode than “clickbait” and “linkbait”
took their place.
We’ve all seen them – headlines that stop at
nothing to get us to drop everything so we
can click on the story and then link to it:
s 32 Freaky Times the World Was Creepy in
the Worst Ways Imaginable
s The Things You Can Find on These 25 Bizarre Islands Seem Too Freaky to Be Real
s She Was Hit by a Car, Struck With a Hammer, Buried...and She STILL Wags Her Tail
And my personal favorite for at least the
past five minutes:
48
The Milken Institute Review
s He Thought He Could Be a Human Anteater, But What Happened Was...OMG
These are all recent examples from ViralNova, a site most readers of this journal have
probably never heard of. Nonetheless, in April
2014 Bloomberg Businessweek declared the
year-old business “one of the defining media
companies of this convulsive era.” What made
it defining was the same thing that once made
Demand Media defining, but with a twist. ViralNova lures millions of people to junk content, not through search but because it can induce people to link to it – mainly on Facebook,
which accounted for 90 percent of the site’s
6.6 million monthly unique visitors.
ViralNova is headed by Scott DeLong, a
31-year-old entrepreneur who lives next to a
cornfield in North Canton, Ohio (pop.
17,500). DeLong doesn’t have a huge news
operation or a vast network of contributors;
he doesn’t need them. All he requires is an eye
for arresting video, the nerve to poach it from
other online sources (many of which have
themselves poached it from someone else), a
savvy way with Facebook, and enough servers
to keep the whole thing from crashing. This
last appears to be his biggest problem.
DeLong is not alone. Upworthy, founded
by the former MoveOn organizer Eli Pariser
and the former Onion editor Peter Koechley,
pioneered the genre in 2012. After six months
online, it could (and did) boast of 8.7 million
uniques, with every post being shared an average of 25,000 times. Ashton Kutcher’s new
site, A+, has 27.5 million monthly uniques in
the United States alone and was recently
hailed on the news site Business Insider as
“one of the most important media companies
in America” – even though, as the writer admitted later in the same sentence, “almost no
one knows it exists.”
And then there’s Emerson Spartz, a 27year-old Chicagoan who created a hugely pop-
ular Harry Potter fan site at 12 and now runs
a potpourri of ad-laden websites that repurpose and repackage content for maximum virality. In a recent New Yorker profile, Spartz
made it clear he thinks originality is for
chumps: it takes too much time, and people
are no more likely to click on the result. So he
and his peers feed off each other while fre-
“influentials,” the highly connected individuals
singled out in Malcolm Gladwell’s best seller
The Tipping Point. Opinion leaders, contrary
to popular opinion, aren’t nearly as critical to
spreading information as ordinary people
who are easily persuadable.
Unruly, a London-based company that
specializes in delivering highly shareable ad
ViralNova lures millions of people to junk content, not
through search but because it can induce people to link
to it — mainly on Facebook, which accounted for 90 percent
of the site’s 6.6 million monthly unique visitors.
quently harvesting the value of information
other people have painstakingly gathered.
All this is possible because we no longer
have to wonder why one story or video goes
viral and another does not. What used to be
viewed as a crapshoot is now almost a science.
Jonah Berger of the Wharton School wrote a
best seller called Contagious after he and a
colleague, Katherine Milkman, analyzed
7,000 New York Times articles that ran in the
fall of 2008. The study showed that a key factor in getting people to email stories to
friends and colleagues was emotional arousal.
In general, people were more likely to email
positive stories than those that generated negative emotions like anger or anxiety. But intensity mattered, too. Stories that sent readers
into a rage were more likely to go viral than
those that evoked emotions that were only
mildly positive, or that were negative in a deactivating way, like sadness. In the book, Berger
cites a handful of other factors as well, among
them social currency, practical usefulness and
good storytelling. Much less important, as
Duncan Watts, a Microsoft researcher and former Columbia professor, has shown, are
videos, offers a far more granular approach to
virality. Its proprietary ShareRank algorithm,
developed from more than 100,000 viewer reactions, gauges the likelihood that a given ad
video will be shared, depending on (among
other things) the psychological response it
generates and the social motivation it appeals
to. Celebrities don’t have much impact, but
friends do; according to co-founder Sarah
Wood, an ad that’s recommended by a friend
is up to 50 percent more likely to trigger a
purchase that an ordinary recommendation.
But while the rate of online video sharing
kept doubling for years, it has recently started
dropping off. The problem, Wood says, is
sheer volume: “We can’t take any more and
pay attention to it.”
Which gets to the heart of the issue. There’s
nothing wrong with virality – it’s how news
and ideas spread in the Internet age. The
problem is with how we measure success and
what kind of behavior those metrics encourage. The focus on uniques and click-throughs
assumes that value online is where it has always been in the media business, in ad inventory. But Web pages can multiply like
Third Quarter 2015
49
bunny rabbits – and the more they do, the less
valuable each ad opportunity becomes. What’s
needed is an injection of scarcity – and “the
only unit of scarcity on the Web,” as Chartbeat’s CEO, Tony Haile, recently told Advertising Age, is time. Herb Simon would approve.
in search of lost time
It follows that, as with any scarce resource,
those who capture users’ time should be able
to charge a premium. The question is, how do
you capture it? Time online is owned by users,
not publishers, so it’s theirs to spend. But if
they spend it on your content, then it should
be yours to monetize – or so argues Haile,
whose company measures traffic for clients
ranging from the Financial Times to Gawker
Media. Haile evangelizes for what he calls the
Attention Web, a Web that looks beyond
clicks, links, uniques, and monthly active
users to actual engagement – something that
no longer has to be guessed at, now that companies like his can capture it on a user-byuser, second-by-second basis.
Last September, Chartbeat became one of
several ad tech companies whose metrics for
time spent and other not-yet-common factors are accredited by the Media Rating Council, the industry body in charge of such things.
Meanwhile, research by Chartbeat and other
companies – among them Google, Microsoft
and Yahoo – has demonstrated that the longer an online display ad is in view, the higher
the brand recall. This challenges the myth
that banner ads don’t work. True, they don’t
get a lot of click-throughs – but neither do
highway billboards or 30-second TV spots or
double-truck magazine spreads. Just because
a banner ad can be clicked on doesn’t mean it
needs to be. Done well, display ads online
should be as effective at brand-building as
television spots have been. The difference is
50
The Milken Institute Review
that, unlike television, the Internet can actually tell advertisers if anybody is looking.
The brand issue is critical, and not just for
advertisers. When a publisher relies on transient traffic from Google or Facebook and
tries to maximize uniques at the expense of
delivering a satisfying experience, it’s never
going to develop brand loyalty. So publishers
have to make a choice: what kind of site do
they want to run?
Upworthy’s founders say they’ve decided
to mend their ways, focusing on engagement
factors like comments and time spent reading.
Instead of optimizing for Facebook shares
and page views, they plan to return to their
original mission, helping people “find important content that is as fun to share as a FAIL
video of some idiot surfing off his roof.”
For people like Upworthy’s Peter Koechley,
Chartbeat’s Tony Haile and Twitter’s/Medium’s Evan Williams, attention metrics are a
make-or-break proposition. “We are in an allout war for attention between the forces of
inanity and the forces of things that actually
matter to society,” Koechley told The Guardian before speaking at its annual Changing
Media Summit last spring. “We feel like people paying attention and being aware of important issues is one of the big roles of media.”
Implicit in this critique is the idea that
short-attention-span behavior is neither the
users’ fault nor the result of shallowness endemic to the Web. Television didn’t have to be
the “vast wasteland” that Newton Minow decried in the 60s, and neither does the Internet.
Wastelands happen when advertisers, publishers and investors allow shallow metrics to
guide their decisions. We reward what we measure, and we get what we reward. No one wants
to turn the Web into some endless educational
TV channel, all uplift and gruel. There’ll be
room for LOL cats and listicles galore. The
question is, what else will we get?
kotryna zukauskaite
th e atte ntion economy
tk
Third Quarter 2015
51